102961



world development report


            BACKGROUND PAPER

            Digital Dividends


            Network Neutrality and
            Private Sector Investment
            Jose Marino Garcia
            World Bank Group
This background paper was prepared for the World Development Report 2016 Digital Dividends. It is made
available here to communicate the results of the Bank’s work to the development community with the least possible
delay. The manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate
to formally-edited texts. The findings, interpretations, and conclusions expressed in this paper do not necessarily
reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent.
The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors,
denominations, and other information shown on any map in this work do not imply any judgment on the part of
The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries.
            NETWORK NEUTRALITY AND PRIVATE SECTOR INVESTMENT


The implications of network neutrality abolition in developing countries are linked to
the problems of lack of affordability and access, and might pose important competition
problems


         Network neutrality and network diversity, a pricing problem

Pricing of Internet connectivity services poses two important regulatory challenges.
The first is which pricing model maximizes social welfare and allows achieving the
most efficient degree of innovation and investment in two inter-dependent markets,
Internet access provision and Internet content supply. The solution might imply the
extraction of benefits through price discrimination to content suppliers from Internet
access providers in what it is known as the network diversity approach. The second
is to discern, whether potential competition problems stemming from vertical
integrations in these two related markets—exclusion from access providers to
content providers—should be solved with ex-ante regulation, in what it is called the
network neutrality approach, or sorted out ex-post by using existing competition
legislation. Economic theory approaches these problems with two main types of
methodologies, two-sided market models analyzing social welfare by looking at cross
externalities between the two markets and models of congestion management
analyzing total welfare in a context of bandwidth scarcity. In developing countries
where affordability and lack of connectivity prevent the access to the Internet, the
network neutrality debate takes a different form and competition problems are
distinct. The goal of this paper is to examine network neutrality economic and
regulatory implications, to analyze how these implications are related to
development goals and suggests policy recommendations.

Network neutrality is a regulatory approach to Internet connectivity pricing where
customers and content providers (CP) pay a price to their respective local Internet
Service Provider (ISP) to access the Internet, ISPs pay Internet Backbone Providers
(IBP) an interconnection price to route and transport traffic and net payments
between IBP’s are normally zero. Under network neutrality all traffic is treated
equally and consumer’s local ISP does not receive any payment from content
providers for the traffic they send through their network; see Figure 1.




                                          1
Network diversity is a pricing model where consumer’s ISP charges the content
provider a fee that depends on the type of content or application. This is a price
discrimination approach that can be implemented in two different ways, by
discriminating on quality of service or on amount of data. Price discrimination based
on quality entails that delay or throughput sensitive services and applications must
pay a fee for guaranteed data delivery, and those content providers not willing to pay
the fee would experiment service degradation. Price discrimination based on amount
of data is based on charging a higher price to content providers that generate higher
amounts of traffic. Additionally, in some variants of the network diversity approach,
consumers might also be charged for different qualities of service. Under network
diversity, traffic is not treated equally anymore and a different fee is charged to each
application or content; see figure 2.




Whereas content and application providers argue that network neutrality is the
cornerstone for an open Internet and has been responsible for the successful
emergence of innovative applications such as voice or video over IP, access providers
respond that investments to expand network access and deploy next generation
technologies are threatened by network neutrality regulations. Extending this logic,
the fulfillment of the externalities linked to the use of broadband might be
endangered by network neutrality, they argue.

From the perspective of the strategic behavior of access providers, two forces are
responsible for their request to abolish the network neutrality principle. Traffic
demand is increasing exponentially, congesting fixed and especially mobile networks,
due to the emergence of new devices such as tablets, smartphones and other devices,
and the increasing usage of multimedia applications, for example Netflix and Google
are responsible for about 50% of downstream traffic in fixed networks during peak
periods in the US.1 There is also a shift underway in the balance between the agents
in the communications value chain, away from the predominance of access providers
towards the increased importance of content providers.

In the developing world where more than half of the population is not online, lack
of access and affordability problems are more important than the network
neutrality debate as it is conceived in developed countries. When the price of
connection is exorbitant it has no sense to discuss if it is convenient to charge

1Source: Financial Times http://www.ft.com/intl/cms/s/0/4610dfe8-7dbf-11e3-95dd-
00144feabdc0.html#axzz38JGLrSoW


                                                2
different prices to different services. If there is not access at all, either because
broadband is not affordable or because the network is simply not available in that
area, how to get users to access is the main problem to solve and network diversity
takes a different form.

The new approach to network diversity that has arisen in some developing countries
consist of content providers paying users broadband data charges provided that they
browse their sites or use their services. Otherwise, users have to pay for the traffic
they consume; see Figure 3. Projects such as Facebook Zero or Google free zone offer
free mobile access to Facebook and Google services, with sites adapted to existing
data speeds. Facebook zero for example is available in 45 countries in Africa and Latin
America and 50 mobile operators have signed an agreement with Facebook to
implement the model.




Increased access and affordability, in this case, is at the expense of reduced
competition in the access, content provision and publicity over the Internet markets.
Regulatory weakness in some developing countries might worsen the problem.

The pricing of Internet services is therefore a competition and a joint innovation and
investment problem. The extraction of benefits from content and application
providers might boost infrastructure investment and innovation but at the same time,
reduce innovation in content generation and innovative services and applications.
The key challenge is finding a pricing model with the correct mix of incentives to boost
competition in both markets so that the social welfare is maximized. There are also
competition problems. Exclusion agreements between Internet access providers and
content and application providers, can be solved with ex-ante regulation, and
therefore address the problem before it happens—network neutrality regulation—
or otherwise, it can be sorted out ex-post by using the existing or additional
competition legislation, once the potential damage to effective competition has
occurred.


                           The regulatory implications

ISPs claim for the abolition of network neutrality. They see content providers as
agents that generate network congestion, profit from the investments they have made



                                          3
in network deployment and do not contribute in the development of the platform they
use to obtain their benefits. In their view, content providers would be extracting
benefits that are necessary to innovate in new access technologies, to invest in
network expansion, and to solve congestion problems created by the exponential
increases in traffic demand. What ISPs silent however, is that the demand for access
provision depends on the demand for content. Without content demand there will not
be access demand. The channels by which content providers might extract value from
ISPs are mainly two:

First, by depriving access providers from the revenues of its legacy services—
telephony, text messages, television etc.—replicating them in the application layer of
IP technology without being subject to the regulatory framework that access
providers are required to comply with. Examples are telephony and voice over IP,
broadcasted television and IP television or SMS and messages over IP. Internet
services that are not directly provided by the access provider are called over the top
services.

Second, access providers undergo increased congestion in their networks, and are
being forced to upgrade infrastructures with fewer revenues due to the extraction of
benefits from over the top applications. Whereas extraction is not an anticompetitive
behavior in itself, it might pose a problem to social welfare if it prevents the
deployment of infrastructures. Some over the top services are generating big amounts
of data traffic pressing networks to the capacity limits as with the case of Netflix and
Google.

To solve the problem ISP’s propose to charge content providers and also users for
increased quality of service—a kind of ‘Premium Internet’—and to charge content
providers with a termination of interconnection fee (Sending Party Network Pays), in
analogy to the regulation of traditional telephone networks.

However, there are reasons to hold the network neutrality approach, in T. Wu and C.
Yoo (2006); Wu considers that the economics of infrastructure in the last mile is the
key element of the network neutrality discussion. Incumbents in a monopoly position
have an incentive to block content and innovative access technologies because it
might threaten their business model. Network neutrality would not solve the
competition problem in the access network, and abandoning the network neutrality
approach would harm the application and content markets in exchange for creating
very limited benefits in the last mile infrastructure problem. Network neutrality is an
effective method to avoid competition problems, such as traffic blocking and
exclusion of competitors. However, whereas this is true in developing countries
where the network diversity approach damage competition this might not be so clear
in developed countries and more research should be undertaken.

According to Wu, without network neutrality an access provider might prevent the
entry of a new innovative application, e.g. VoIP, but expecting increases in innovation
and investment in the access network through the abolition of network neutrality is
an unrealistic assumption about incumbent behavior. Access providers will have an
incentive to distort competition to maintain the current level of scarcity in the last
mile. A welfare superior and more plausible solution to solve the problem might be


                                          4
government intervention, by subsidizing network deployment, or building the
infrastructure itself. Examples of this proposal are roads in the US, fixed broadband
in many Asian countries or the subsidies offered by Mexico to deploy the mobile
access network using the 700 MHz band. Doing nothing waiting for market or
technology to overcome the last mile infrastructure problem is also an option, but a
risky one since by doing nothing the outcome might be that nothing happens.

Proponents of network diversity however, argue that network neutrality regulation,
with the exception of some cases of social production, is unnecessary because the
problems that is trying to address or does not exist or can be solved with existing ex-
post regulation. Whereas this can be supported in the case of developed countries it
can be easily challenged for developing countries.

C. Hemphill (2008) enumerates a set of competition problems that network diversity
might pose, and analyzes whether network neutrality is a solution for them. The first,
is the case of exclusion accomplished by contract. It occurs when a content provider
linked to an access provider agree to exclude or degrade the service of a competing
provider. In this case, network neutrality would be unnecessary since antitrust-law
already deals with the problem, as the Microsoft-Netscape exclusion case proved.
However, Microsoft’s example also shows how lengthy the process is and how this
can enable the monopolist to keep its market power. Furthermore, the vertical
integration problem is already happening in developing countries where content
providers pay operators to provide free access to their applications, creating a barrier
to the use of other contents as we have seen before.

The second occurs in the event of exclusion accomplished through refusal to deal, e.g.
the case of AT&T vs. Vonage, but existing fragmentation of access provision and
antitrust prohibitions might solve this competition problem. However, in some cases
fragmentation is not enough. There is also an additional problem; the exclusion of
social production. This situation occurs when an access provider is able to collect
profits from private production but not from social production. In this case the access
provider would have an incentive to exclude the social producer, even if social
production is more efficient than market production. This is the case of Wikipedia.

Finally, the extraction problem takes place when an access provider charges a content
provider for access. In this situation content provider’s profit falls and the incentive
to innovate and invest in content production is reduced. However, not only content
innovation must be taken into account but also infrastructure innovation.

In T. Wu and C. Yoo et al (2006), Yoo considers that in the absence of demonstrated
economic effect, practices should not be categorically prohibited. Practices that are
sometimes harmful and sometimes beneficial should be subject to the “rule of reason”
that entails to analyze case by case to evaluate the potential competitive harm of an
action and to allow going forward until a concrete harm to competition has been
proved. Regulatory intervention should be limited to prohibit vertically integrated
owners from blocking content and applications that compete directly with their own
offerings. Yoo maintains that Network neutrality threatens investment and
innovation in the last mile networks by preserving high market entry conditions and
therefore harming competition. Leaving behind the network neutrality approach


                                          5
might open competition in the last mile networks allowing for three different kinds
of content networks, a network for traditional Internet applications, such as email and
Web browsing, a network for e-commerce with security features, and a network for
time sensitive applications such as streaming, media and telephony, all of them with
a different pricing structure.

The fall in the cost of establishing last mile networks due to the emergence of
spectrum-based solutions will change the economics of the last mile infrastructure,
making competition feasible and the market contestable. In this new context,
promoting competition in the complementary services through network neutrality is
no longer the better policy response. The first best policy is promoting competition
in the last mile.

Countries position towards the problem is diverse. In some countries such as Chile
and the Netherlands the law enforces network neutrality and prevents the use of deep
packet inspection, In the US, regulation prohibits the existence of paid prioritization
of any kind and bans the existence of fast lanes, but the European Commission
revision of the open internet regulation has allowed for a different treatment of
specialized services such as IPTV, automated driving or healthcare services provided
that these services do not harm the availability and quality of the open Internet
Access.

Finally, political manipulation of the Internet and a negative effect on human rights
have also been related to network diversity but although it is true that network
diversity might facilitate the manipulation of information by making filtering
technologies available, information censorship and manipulation and the network
neutrality problem are different issues. Censorship happens even with network
neutrality regulation, and the same mechanisms to guarantee freedom of opinion that
are in operation should prevent censorship with network diversity.



                The economics of network neutrality and diversity

Economic analysis uses several methodologies to look into the effects of pricing of
connectivity services on total welfare; two are the more prevalent; two sided market
and congestion management models.

Two-sided market models consider two groups of agents, the sides, interacting,
through a platform. The benefits that each of the agents obtain from joining the
platform depend on the size of the group to which the agent belongs, and the other
side group’s size. The benefits are known as network externalities. There are
externalities related to the size of each group and cross-externalities between groups.
This means that the more users connect to the Internet the greater is the utility that
a user obtains from connecting to the network. The reasons are straightforward, e.g.
the greater the number of e-mail users the higher possible destinations for an e-mail.
Cross externalities also exist, the more contents are available, the greater is the utility
for consumers to connect, and the more consumers, the greater the market for
content providers.


                                           6
In the two-sided market approach, the key parameter to look at is the relative value
of the network effects between the two sides of the market; the content-supply side
and the consumer-demand side.




The determinants of equilibrium prices between agents of a two-sided platform are
the magnitude of the cross-group externalities, whether platform fees are flat-rate or
payments per amount of traffic, and whether the agents have multi-homing
possibilities. Where multi-homing refers to the possibility for a user or content
provider to connect to the Internet using different ISPs at the same time, Armstrong
(2006).

The social optimum and the number of consumers and content providers that would
exist in a market depends on the value of cross-group externalities in both directions,
the price that the ISP charge to consumers and the fee that the platform charges to
content providers in the network diversity approach, for most values of these
parameters, total surplus is higher with network neutrality; N. Economides and J. Tag
(2012).

Congestion models are more heterogeneous and analyze bandwidth scarcity to
disentangle the effect of internet access pricing on welfare, private investment and
innovation.

Network diversity can avoid congestion by creating a price for a preferential access
to the Internet, but hurt content providers investments; Cheng et al. (2007).
Prioritization would make consumers to switch to the content provider that bought
prioritization to the ISP, improving productive efficiency but at the same time,
consumer utility losses will happen because they would be deprived from the use of
some of their preferred services; Choi and Kim (2010). The conditions for network
neutrality to be a total welfare superior option compared to network diversity in the
presence of congestion, occurs when the elasticity of content demand with respect to
transmission time does not increase with households’ delay sensitivity for the
content; Economides and Hermalin (2012).

The effects of access pricing on social welfare depends on the particular values of the
parameters in all models, however, there is no empirical research of the values for
most of these parameters. Additionally, any of the models incorporate the
externalities of Internet access to the economy as a whole in terms of increased



                                          7
productivity and employment. Only the externalities between sides in the two-sided
market models are considered.

What is the different impact of network diversity in developing countries and in
developed economies is still a question that remains unapproached in economic
analysis and needs to be studied. An important element in the analysis must be the
different patterns of traffic demand in developing and developed countries and the
impact of network capacity in shaping demand. For example, whereas in North
America the main contributor to traffic demand is streamed audio and video (real
time entertainment) with almost 60% share of the total traffic, in Africa, Web
browsing is as important as video on demand, and both accounts for around 25% of
total traffic; see Figure 5.

         Figure 5: Peak traffic composition in North America and Africa




  Source: Sanvine Analytics. Available at: https://www.sandvine.com/downloads/general/global-
            internet-phenomena/2014/1h-2014-global-internet-phenomena-report.pdf

Empirical literature is scarce and diverse from the point of view of the methodology
used: R. Clarke (2009) uses a cost model to argue that network costs levels would
exceed by an order of magnitude today’s levels and therefore neutrality is not be a
feasible option. Traffic demand will exceed supply and networks will not be able to
satisfy video and other high bandwidth contents with neutrality.


                                Policy recommendations

Network diversity is not a market demand, and network neutrality in developed
countries seems to be a solution for non-existing competition problems. In the case of
developing countries network diversity is a trade-off between affordability and
competition.

Whereas literature is focused on whether competition problems between access
providers and content providers should be solved ex-ante or ex-post, more attention
should be given to the value proposition that a differentiated Internet service would


                                             8
offer to customers. New technological developments suggest that network diversity
is not a market demand, since technology has reduced the necessity of differentiated
Internet services. Protocols that control congestion at the end point, assume no
network support to quality of service, and use packet losses as a signal of congestion
to control self-sending rate, reducing network congestion, improving jitter and
yielding a fair allocation of network elements. Internet Exchange Points and peering
are also reducing the traffic to deliver to the Internet backhaul and backbone,
reducing per bit delivery cost, and improving routing efficiency, packet losses and
latency. Content delivery networks such as Akamai are also helping to reduce the
need of end-to-end solutions. After years of innovation in the effort-less Internet it is
maybe too late for access providers to suggest a dedicated path quality of service
solution for congestion. The consequence is that network diversity and differentiated
services are less necessary to provide low latency and/or low jitter services such as
VoIP or video on demand. Only very critical jitter and latency applications such as
telemedicine might need a dedicated path and a reserved capacity in each node to
succeed.

The existing experience in telephone markets, shows that models where calling
network party pays, have not achieved increased competition and investment in the
access network, on the contrary, it indicates that incumbents will behave restricting
competition in the last mile. Substituting the existing network neutrality, for a calling
party network pays approach, might not increase investment in the access network
and therefore benefits from departing from network neutrality would not
compensate for the losses—innovation in the content and application industries.

However, traffic growth must be monitored, since increased congestion and the
subsequent demand for network deployment at a higher pace than in the past might
imply a fall in the present value of the gross profits per user that access providers
obtain from building and operating the network. This fall would prevent investments
in network deployment and as a consequence a social under-provision of broadband.
Investments in network upgrades will only be undertaken up to the point that the
present value of the necessary costs to satisfying the demand equals the present value
of the profits to be obtained. For this reason, increases in traffic demand might make
network diversity a social welfare superior option.

Empirical evidence of anticompetitive behavior of access providers is weak in
developed countries. Over a decade, the FCC only found four related cases of
anticompetitive behavior, suggesting that the ex-post approach to potential
competition problems of network diversity might be adequate. However Microsoft-
Netscape antitrust case can be seen as an example of how long is the ex-post process
and how the delay can allow the monopolist to keep its market power.

In developing countries, however, access and affordability are at the expense of
reduced competition in the access, content provision and publicity over the Internet
markets. Regulatory weakness in some developing countries might worsen the
problem. If a subsidy to consumers is to be allowed, regulatory measures such as the
guarantee that all mobile operators are offered the same deal must be adopted.




                                           9
A subsidy to consumers allows content providers to create a monopoly in related
markets to extract monopoly rents. The agreement between Facebook or Google and
a mobile operator is an exclusion deal that prevents other content providers to access
the market by increasing entry barriers. The agreement also cut down competition in
the publicity over the Internet market, since the options for advertising are drastically
reduced. Content providers might cross-subsidize developing countries
advertisement markets with the developed countries revenues to gain monopoly
control of the publicity over the Internet market. Furthermore, if Facebook or Google
are not mandated to offer the same agreement to all existing mobile operators, a
competition problem in the mobile access market will appear.



References


Wu, T., Yoo, C. 2006. Keeping Internet neutral? : Tim Wu and Christofer Yoo Debate.
Federal Communications law journal. Vol 59. Number 3.

Hemphill, C., 2008. Network Neutrality and the false promise of zero price regulation.
Yale Journal on Regulation. Vol 25:2, 2008.

Armstrong, M., 2006. Competition in two-sided markets. RAND Journal of Economics
37, 668–691.

Economides, N. and Tag, J. “Net Neutrality on the Internet: A Two-Sided Market
Analysis.” Information Economics and Policy, Vol. 24 (2012), pp. 91–104.

Hermalin, B., Katz, M.L., 2007. The economics of product-line restrictions with an
application to the network neutrality debate. Information Economics and Policy 19,
215–248

Chen, M.K., Nalebuff, B., 2007. One-Way Essential Complements. Mimeo. Yale University
(April).

Choi, J.P., Kim, B.C., 2010. Net neutrality and investment incentives. RAND Journal of
Economics 41, 446–471.

Economides, N., Hermalin, B., 2012. The Economics of Network Neutrality.
RAND Journal of Economics.

Clarke, R. 2009. Costs of neutral/unmanaged IP networks. Review of network
economics.




                                          10